Google

keyword contextWhen discussing search results, I often spend far too much time talking simply about ‘keywords’ without explaining the contextual factors that influence the results Google return for a given search query.  In this post I’m going to attempt to explain why we shouldn’t focus solely on keywords, and why we also need to consider the context and user intent of a search query to understand how we can best serve them.

Search Queries as Keywords

Traditionally, we have thought of search queries using a keyword-focused model, which still makes sense to an extent. However, search engines are changing – they’re getting much better at understanding user intent whenever a search is carried out.  Search engines such as Google will specifically look to return results that meet the expectations of the user.  Search engines incorporate semantics to grasp the meaning of a search query, as they want to return what they feel are the most appropriate search results to help answer the query. This all sounds simple in theory, but there are hundreds of factors that will need to be taken into account before Google can fully understand the intent of a search query.  For example, some questions Google might ask about a query for “nottingham tram“, are as follows:

  • Is this a user looking for news about the Nottingham tram network?
  • Is this someone looking for a history of the Nottingham tram network?
  • Is this someone looking for a list/map of all the Nottingham tram stops?
  • Is this someone looking for a tram they may be able to catch nearby?

Search Queries as Keywords in Context

There are clearly plenty of possible situations in which Google will need to assess the intent of the search query before deciding on which results to return, and it will sometimes prove difficult for Google to determine what the user is actually looking for (especially when broad search queries are used as opposed to longer tail phrases). It is therefore important that we understand that the keyword(s) we type in do not make up the entire search query; meaning the keywords themselves are not the only thing Google will use to answer a search query.

A search query actually looks more like this to Google:

keyword context

 

The contextual factor I have used above is location, when in reality there are many more factors Google take into account when trying to establish the context in which a search query occurs, such as:

  • The search history of the user, which is why Google are able to deliver personalised search results
  • Location of the user: depending on where the user is based, the search engine will serve results more suited to the area the user is based
  • Device the user is using at the time the search query was made
  • Relationships between a previously stored data (named terms or entities)
  • The characteristics of the keywords used: spelling, grammar etc.
  • and many more.

Search engines such as Google are now becoming much better at connecting the contextual factors above, with the explicit intent of the user (the keywords used in the search query).

Local vs. National Results

Location is perhaps the most obvious contextual factor Google uses to serve personalised results for a search query.  Location is a particularly important factor for certain keywords, which Google respond to by serving local businesses or attractions to make their results more useful.  For example, if you try searching Google for the keyword “electrician” you should only see local electricians appear in the results – even if you’d been living under a rock your entire life, and in fact wanted to see results that define the job of an electrician.  Google decide whether their search results should be localised or not based on the keywords used.  In some cases, it makes more sense for them to serve local results – for example, something such as “Chinese restaurants” but they’ll serve  national results for something such as “mortgage providers”.  If your target keywords tend to serve local results, then your website is only likely to be served in results for search queries carried out within your local area.

What does this mean for keywords?

For as long as there are search engines, there will be keywords.  However, our approach to keywords is changing.

A few years ago, it would have been unthinkable for a small restaurant to rank on the first page of the search results for the terms “restaurant” or “pancakes”. That was until the Venice update and suddenly local businesses were able to rank highly for these keywords within certain contexts.

In addition to this, if we consider the move towards the knowledge graph, and Google’s shift towards contextual search, it’s clear that focusing on keywords alone would be ludicrous moving forward.  Understanding the context of search queries is something Google are clearly already getting better at, and we need to ensure that we try to understand how Google decipher the intent of various keywords to ensure we’re able to create content that is more likely to be served for those keywords, in the desired context.

Search Entities

To better understand this, it may be useful to look at how Google make the connection between the words used in a search query and the intent of the user.  In order for Google to display the most relevant results, they use search entities.  Entities are the people/things/places/subjects that words or phrases could be referring to.  For instance, a user could type in the word ‘bear’ as a search query.  Google will then attempt to match results to this search query by serving results connected to an entity that is associated to the word/s typed in.  To return the best answer, they use probability measures (amongst other factors), as explained in greater detail here.

Now, let’s go back to the search query in question:

bear uery

 

As I’ve mentioned, ‘bear’ is a very broad phrase that could refer to a whole range of separate entities.  In this case, Google are suggesting I might be searching for a person (bear grylls).  To explain the entity model further, let’s take a look at what else I could be referring to with a simple search for ‘bear‘:

 

query entity model

 

In the end, I went ahead and conducted a search for the word ‘bear’, and received results based largely around the animal, as Google believe that this is the most probable entity that people will be searching for when using the word ‘bear‘:

bear

Google have decided via a process of word-sense disambiguation, probability measures and other factors (as discussed by Bill Slawski), that most people searching for the word ‘bear’, mean to look for the animal. However, it is still important to note that Google do their best to provide alternative suggestions within their results for anyone who may have been searching for alternative meanings of the word in question (see ‘bear (gay culture)‘ link underneath the Wikipedia listing, for instance).

Are keywords still important?

In short, yes. As long as people are performing language driven searches via either text or spoken word (think siri), keywords are obviously going to be important. Whatever a user explicitly enters as part of their search query is clearly always going to be important.  However, we need to stop looking at keywords as just that and starting looking at the characteristics of search queries. A search query contains both explicit and implicit (contextual) aspects, which I’ve attempted to explain thus far.

Capitalising on Semantic Search

Identifying Synonyms and Keyword Variations

Instead of repeating a set of target keywords across your website, you need to aim to uncover the questions people may be asking when looking for products and services similar to those you’re able to offer (your target keywords). Identifying these kinds of questions/long-tail queries will give you a hit list of search-friendly pages/articles to write around the topics you discuss/sell/offer on your website, and a much broader reach within the SERPs.  The aim of this activity is to build a bank of rich content that is important in multiple contexts for search queries related to your existing content.

Utilising Schema

For much of this post I have discussed the ways in which Google are able to understand the semantics behind various search queries. It is important to note that you can help search engines understand the content on your website by using semantic markup available at schema.org.

Schema.org allows webmasters to find the reference needed to semantically markup their content appropriately.  As stated on their website:

On-page markup enables search engines to understand the information on web pages and provide richer search results in order to make it easier for users to find relevant information on the web.

Final thoughts

Bearing in mind the points I’ve just covered, there is an obvious need to move away from the purely keyword focused model of search marketing, and start to take note of the variety of contexts involved in a given search query (or set of search queries).  When we look to point out potential sources of content, analysing contexts by creating “context personas” is going to become increasingly important.  There have always been users with a variety of different needs (in various situations) using the same/similar keywords, the difference now is that Google are getting much better at promoting results that match the context of each search query.  We therefore need to ensure we’re creating content to meet the needs of users across various contexts.

Finally, the way we report to our clients (or management) needs to begin to change in some instances. Reporting on keyword data alone is going to become less worthwhile (especially considering the rise in keyword ‘not provided’), and it’s time to start educating our clients so that they understand this shift in advance.

Leave a Reply

Your email address will not be published. Required fields are marked *